Detection and Analysis on Deep Convective Clouds in a Frontal Cyclone Using Multispectral Remote Sensing Data

Detection and analysis on deep convective clouds in a frontal cyclone using NOAA-16/AMSU-B and GOES-9 data are investigated. A series of detection algorithms and discrimination are adopted, including the microwave brightness temperatures detection from the two window channels, water vapor channel microwave brightness differences identification based on the NOAA-16/AM SU-B data, infrared brightness thresholds detection of cloud top temperatures, the water vapor and infrared window temperature differences determination, and the classification of cumulonimbus clouds correlating with deep convective clouds applying with infrared/water vapor spectral features. These methods are validated by overlaying surface conventional data on the results and comparing them. The results show that microwave brightness temperatures from window channels can discriminate deep convective clouds effectively, while the brightness temperatures of 89 GHz are affected by surface features and the cold water surfaces are mistaken to convective clouds. The brightness temperatures of 150 GHz are just slightly influenced by surface characteristics, so the detection areas are coincident with those from water vapor channel microwave brightness differences identification, which can identify the deep convective clouds well and depend on the thresholds less. As to GOES-9, different infrared brightness thresholds bring about significant detection differences. Single thresholds are applicable to the local areas and the thresholds applicable to global regions should vary in spatial and temporal scales. The water vapor and infrared window temperature differences can detect convective regions well, while the determination areas are smaller. The stepwise cluster can identify cumulonimbus clouds correlating with deep convective clouds by means of infrared/water vapor spectral features, which can classify clouds objectively by combining image and pattern recognition. The detection areas are coincident with NOAA-16/AM SU-B detection areas, and the surface conventional data can validate the results, including hazards weather and cumulonimbus clouds.